1. Field
Example embodiments of the present patent application generally relate to an image retrieval apparatus, an image retrieving method, and a storage medium thereof, and particularly to an image retrieval apparatus included in an image forming apparatus or an image processing apparatus so as to retrieve an image similar to specified image data, an image retrieving method used in the image retrieval apparatus, and a storage medium storing a machine-readable program for causing the image retrieval apparatus to perform the image retrieving method.
2. Discussion of the Related Art
Ordinary or general image retrieval apparatuses have a configuration to input and store image data and retrieve the stored image data when needed. Further, such general image retrieval apparatuses may have a function that can retrieve similar images.
When such general image retrieval apparatuses include a data display unit that is not large enough to display the stored image data or the like, a user may confirm the stored image data, for example, by checking such image data information contents as names, dates, registrants, file sizes, etc. listed up and displayed via the data display unit. The user may also confirm the stored image data through a reduced image, e.g., a thumbnail-size image of the entire image data that is selected intuitively. However, when such reduced images are used, the user cannot reliably distinguish between similar though not identical images.
For example, FIG. 1 shows a display screen using a related-art technique in which reduced images are displayed.
In FIG. 1, a reduced image of an entire original document 10 and a reduced image of an entire original document 10a that resembles the original document 10 are displayed side by side on the image display unit 2 or a display screen. As shown in FIG. 1, the reduced image of the original document 10 and the reduced image of the original document 10a have the same layout. That is, from the reduced images one cannot tell the differences between the original document 10 and the original document 10a. Therefore, details of the original document 10 and the original document 10a may need to be examined to find if there is any difference between them.
In using such a reduced image, there are some techniques to distinguish one image from another image of similar layout.
In one technique, for example, a reduced image preparing part reads electronic data such as image data, documents prepared by a word processor, or data created by a spreadsheet program, and then prepares the reduced images. A difference determining part then finds a degree of similarity between each of the plurality of prepared reduced images. When the degree of similarity is larger than a given threshold value, the difference image preparing part cuts areas of different portions out of the similar images and an image displaying part displays the portions to aid in distinguishing the images.
In another technique, a document sorting unit extracts respective layouts of a plurality of documents and sorts the documents into categories based on the layouts, a dissimilarity extraction unit extracts dissimilar portions in each document, and a dissimilarity display unit displays only the portions extracted by the dissimilarity extraction unit. Further, a similarity extraction unit extracts similar portions in each document, and a similar portion display unit displays only the portions extracted by the similarity extraction unit.
The forgoing techniques display images of a lesser degree of similarity out of the multiple partial images of an original image when the reduced images are similar to each other. Thus, by displaying such images of lesser similarity, similar images can be distinguished even when the reduced images of the respective entire image data are similar to each other and difficult to distinguish.
Further, in another technique, a document file acquisition unit acquires a document file, a reduced image generation unit generates a reduced image of an image of the document file acquired by the document file acquisition unit, a processed area determination unit determines a processed area in the image of the document file when the reduced image displayed on a display is specified with a pointer, an extracting unit for extracting information in the processed area extracts information in the processed area from the document file, and a display unit for the information in the processed area displays the information in the processed area on the display in a specified size. With this configuration, when an operator specifies a focused portion of the reduced image displayed on the display, an area in the vicinity of the specified portion is extracted from the original image data of the reduced data. Then, information is extracted from the area so that the extracted information can be output to display. Accordingly, by displaying information of the focused portion of the reduced image, different image data having similar reduced images can be distinguished.
However, when multiple results of retrieval are displayed in the first technique and the second technique, the multiple results of retrieval are displayed as reduced images due to a limitation in size of a display unit. With the reduced images, it is difficult for a user to learn respective features of the images. Therefore, to display a distinctive feature area image of the retrieval results, an image having a least similar image with respect to a partial image in other images is displayed.
However, such a distinctive feature area image according to the first technique and the second technique is an image having a least similarity with respect to the partial image in the other image displayed as the retrieval results. The distinctive feature image is simply an image that is related to the other images, and may not be the distinctive feature area of the original image. Further, when many images are displayed as the retrieval results, the amount of calculation required may slow the processing speed.
Further still, it is not effective to display the information of the focused portion of the reduced image without knowing in advance which portions can be different or having enough experience to predict the difference. When the operator lacks such information or experience, the operator may need to specify the focused portion randomly, which is not efficient.
Finally, as described above, after the operator specifies the focused portion, the area in the vicinity of the specified portion is extracted from the original image data of the reduced data, and the operator must wait while the above-described operation is processed. When a more complex operation is performed, the operator must wait longer than for a simple operation.